Proposed new algorithm for calculation of root switch for multicast
spanning tree. Only edge switches(those connected to hosts) and
switches - multicast members themselves are involved in root calculation.
This gives improvement, especially on large fabrics, since number of
switches usually much
Sasha Khapyorsky wrote:
On 10:09 Sun 01 Nov , Eli Dorfman (Voltaire) wrote:
Report SA MAD Class specific status.
Fixes wrong error report for SA query status.
I agree with patch, but one comment is below.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
Report SA MAD Class specific status.
In addition to SM status.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
infiniband-diags/src/saquery.c | 45 ---
1 files changed, 41 insertions(+), 4 deletions(-)
diff --git a/infiniband-diags/src/saquery.c
Hi Sasha,
OSM can dump SA DB, but it is done every heavy sweep, and
only when running with high verbosity, which cannot be used
in production.
OSM can also load SA DB from file, but then it also stays
in this static SA configuration.
The following patch series improves SA DB persistence:
1/3:
Added option to load SA DB once: 'sa_db_load_once'.
This will cause OSM to load SA DB once during first master
heavy sweep, and then OSM will move to the usual SA mode.
The option is not exposed through OSM command line,
but only through options file.
Signed-off-by: Yevgeny Kliteynik
Added option to dump SA DB at every sweep (both heavy
and light): 'sa_db_dump'.
The option is not exposed through OSM command line,
but only through options file.
Signed-off-by: Yevgeny Kliteynik klit...@dev.mellanox.co.il
---
opensm/include/opensm/osm_subnet.h |5 +
Hi Eli,
On 13:40 Fri 30 Oct , Eli Dorfman wrote:
From: Eli Dorfman e...@voltaire.com
Please add descriptive change log. It is hard (for me) to just remember
an issue in all details.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
opensm/opensm/osm_sa_path_record.c | 38
On Thu, Oct 22, 2009 at 11:01 AM, Sasha Khapyorsky sas...@voltaire.com wrote:
On 08:50 Wed 21 Oct , Hal Rosenstock wrote:
Also, cosmetic change to OSM_LOG message
Please don't mix.
I used Unicast Cache (with upper case starting letters) as for a name.
A nit but any reason for that one
Return single PathRecord for SubnAdmGet with DGID/SGID wildcarded
Instead of iterating over all ports in the fabric and returning an error
(TOO_MANY_RECORDS),
when SGID and/or DGID are wild carded return only single PathRecord.
Signed-off-by: Eli Dorfman e...@voltaire.com
---
On 11:31 Mon 02 Nov , Hal Rosenstock wrote:
A nit but any reason for that one message being inconsistent (in terms
of case) with the other log messages there ?
Other messages are debug printouts and cannot be compared.
Sasha
--
To unsubscribe from this list: send the line unsubscribe
On 11:23 Mon 02 Nov , Hal Rosenstock wrote:
On Mon, Nov 2, 2009 at 9:03 AM, Sasha Khapyorsky sas...@voltaire.com wrote:
On 18:16 Fri 30 Oct , Hal Rosenstock wrote:
Because a PMA implementation which follows the IBA recommendation will
reject this
Hmm, where did you find such
On Mon, Nov 2, 2009 at 12:13 PM, Sasha Khapyorsky sas...@voltaire.com wrote:
On 11:23 Mon 02 Nov , Hal Rosenstock wrote:
On Mon, Nov 2, 2009 at 9:03 AM, Sasha Khapyorsky sas...@voltaire.com wrote:
On 18:16 Fri 30 Oct , Hal Rosenstock wrote:
Because a PMA implementation which
On Mon, Nov 2, 2009 at 12:01 PM, Sasha Khapyorsky sas...@voltaire.com wrote:
On 11:31 Mon 02 Nov , Hal Rosenstock wrote:
A nit but any reason for that one message being inconsistent (in terms
of case) with the other log messages there ?
Other messages are debug printouts and cannot be
Sean, using unmapped multicast addresses I see that a different broacast group
is
created by the SM such that mckey doesn't manage to join the ipv4 broadcast
group
$ ./mckey -M ff12:401b::0:0:0:: -b 10.10.5.62 -p 0x2
Unmapped multicast groups only support the case where the SA has
On 17:59 Mon 02 Nov , elid wrote:
Return single PathRecord for SubnAdmGet with DGID/SGID wildcarded
Instead of iterating over all ports in the fabric and returning an error
(TOO_MANY_RECORDS),
when SGID and/or DGID are wild carded return only single PathRecord.
Signed-off-by: Eli
On 12:18 Mon 02 Nov , Hal Rosenstock wrote:
Yet another (likely even more efficient) approach would be memset()ing
MFTs in realloc function above requested mlid_offset, then we will be
able to remove osm_mcast_tbl_clear_mlid() completely.
Isn't mlid clearing done on a per mlid basis
On 12:44 Mon 02 Nov , Hal Rosenstock wrote:
clear_mlid could be implemented with memset rather than loop.
Yes, memset() would be better.
Sasha
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info
Is there any way to recreate the roll-up digests that we used to get on this
mailing list?
--
To unsubscribe from this list: send the line unsubscribe linux-rdma in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
On Sun, Nov 01, 2009 at 11:31:22AM +0200, Or Gerlitz wrote:
Sean, using unmapped multicast addresses I see that a different broacast
group is
created by the SM such that mckey doesn't manage to join the ipv4 broadcast
group
$ ./mckey -M ff12:401b::0:0:0:: -b 10.10.5.62 -p 0x2
Before enhancing the rdma-cm to support the full feature set of the IB
CM, something which I personally don't see the actual need for (but I
will be happy to get educated what applications will or can migrate to
rdma-cm once this is implemented), how about trying to allow for reduced
QoS scheme
Should be IB_MCAST_POSITION_MAX + 1 rather than just IB_MCAST_POSITION_MAX
Signed-off-by: Hal Rosenstock hal.rosenst...@gmail.com
---
diff --git a/opensm/include/opensm/osm_mcast_tbl.h
b/opensm/include/opensm/osm_mcast_tbl.h
index 6d3f083..0745b5b 100644
---
Hey Sasha,
After talking to Ira about it, we think it's best remove
ibnd_update_node.
A) It's not being used.
B) It probably not implemented properly
C) Some of Ira's original plans for its use require more API functions
to be added, which of course isn't done right now.
So for now, it seems
Hey Sasha,
This splits out some scan specific data from ibnd_node_t that doesn't
need to be in the public struct.
Al
--
Albert Chu
ch...@llnl.gov
Computer Scientist
High Performance Systems Division
Lawrence Livermore National Laboratory
From: Albert Chu ch...@llnl.gov
Date: Thu, 29 Oct 2009
Hi Sasha,
Oops. I forgot to free the newly created memory. Here's a new patch.
Al
On Mon, 2009-11-02 at 11:33 -0800, Al Chu wrote:
Hey Sasha,
This splits out some scan specific data from ibnd_node_t that doesn't
need to be in the public struct.
Al
--
Albert Chu
ch...@llnl.gov
Pradeep Satyanarayana wrote:
This crash was originally reported against Rhel5.4. However, one can recreate this crash quite easily in OFED-1.5 too.
I understand that you get the crash when working with the RHEL5.4
bonding driver, correct? does it happen only with IPoIB devices acting
as the
25 matches
Mail list logo
